Hyderabad: The field of artificial intelligence is entering a transformative phase, driving significant advancements in biology, genomics, and materials science. According to Pushmeet Kohli, Vice-President of Science at Google DeepMind (UK), the objective is to develop AI ethically to enhance human welfare by applying self-learning systems to pressing scientific challenges.
After showcasing their superhuman capabilities in games with AlphaGo, DeepMind redirected its focus toward scientific endeavors. A landmark achievement was AlphaFold, an innovative AI system that predicts the three-dimensional structure of proteins based on their amino acid sequences—a puzzle that had long perplexed scientists, Kohli explained in his plenary session at BioAsia.
Launched in 2020, AlphaFold demonstrated remarkable accuracy in predicting protein structures. The system has gone on to forecast structures for nearly all known proteins and is now utilized by over three million researchers worldwide. This pioneering work has received international acclaim, culminating in a Nobel Prize in Chemistry awarded to members of the team.
The influence of AlphaFold is far-reaching, impacting drug discovery, antimicrobial resistance research, and enzyme development. Its upcoming iteration, AlphaFold 3, will broaden its predictions to include interactions among proteins, small molecules, and nucleic acids, offering deeper insights into cellular biology, Kohli noted.
DeepMind is also making strides in genomics research. Kohli highlighted that while the Human Genome Project laid the groundwork for reading genetic code, comprehending its implications poses a significant challenge. Tools like AlphaMissense aid in identifying potentially harmful genetic mutations, which is crucial for diagnosing rare diseases, whereas AlphaGenome models gene expression and various biological effects.
Currently, the company is working on more generalized “AI Co-Scientist” systems that can generate and refine scientific hypotheses. In one instance, their system derived results in a matter of days that had previously taken researchers almost a decade to accomplish, Kohli mentioned. While AI has the potential to accelerate scientific discovery, he emphasized the importance of utilizing it responsibly and being aware of its limitations.
In conclusion, the integration of AI in scientific research holds promise for unprecedented advancements across various fields. As tools like AlphaFold and future AI Co-Scientist models evolve, they not only enhance our understanding of biological systems but also underscore the need for ethical considerations in leveraging this powerful technology.
